Validated administrative data based ICD-10 algorithms for chronic conditions: A systematic review.

OBJECTIVE: This systematic review aimed to identify ICD-10 based validated algorithms for chronic conditions using health administrative data. METHODS: A comprehensive systematic literature search using Ovid MEDLINE, Embase, PsycINFO, Web of Science and CINAHL was performed to identify studies, published between 1983 and May 2023, on validated algorithms for chronic conditions using administrative health data. Two reviewers independently screened titles and abstracts and reviewed full text of selected studies to complete data extraction. A third reviewer resolved conflicts arising at the screening or study selection stages. The primary outcome was validated studies of ICD-10 based algorithms with both sensitivity and PPV of ≥70 %. Studies with either sensitivity or PPV <70 % were included as secondary outcomes. RESULTS: Overall, the search identified 1686 studies of which 54 met the inclusion criteria. Combining a previously published literature search, a total of 61 studies were included for data extraction. The study identified 40 chronic conditions with high validity and 22 conditions with moderate validity. The validated algorithms were based on administrative data from different countries including Canada, USA, Australia, Japan, France, South Korea, and Taiwan. The algorithms identified included several types of cancers, cardiovascular conditions, kidney diseases, gastrointestinal disorders, and peripheral vascular diseases, amongst others. CONCLUSION: With ICD-10 prominently used across the world, this up-to-date systematic review can prove to be a helpful resource for research and surveillance initiatives using administrative health data for identifying chronic conditions.

Positive predictive value and sensitivity of cancer-associated venous thromboembolism diagnoses in the Danish National Patient Register.

INTRODUCTION: Hospital discharge diagnoses from administrative registries are frequently used in studies of cancer-associated venous thromboembolism, but the validity of International Classification of Diseases (ICD) codes for identifying such events is unknown. MATERIALS AND METHODS: Using patient samples from the Danish National Patient Register, we calculated positive predictive values (PPV), i.e., the proportion of registered ICD codes, which could be confirmed after manual search of the electronic health record. Sensitivity was estimated in a sample of patients with imaging-verified venous thromboembolism but without prior knowledge about their ICD coding status. Sensitivity was calculated as the proportion of these patients, who were discharged with an ICD code for venous thromboembolism. RESULTS: The overall PPV of an ICD-10 diagnosis of cancer-associated venous thromboembolism was 75.9 % (95 % confidence interval 71.3-80.0). In subgroups, the PPV was particularly low for recurrent venous thromboembolism (44.2 %), diagnoses in a secondary position (55.7 %), outpatient diagnoses (65.3 %), and diagnoses given at surgical (66.7 %), emergency wards (48.4 %), or via hospices/palliative teams (0 %). The overall sensitivity was 68 %, meaning 32 % of patients with cancer diagnosed in hospital with venous thromboembolism were discharged without any registered ICD code for venous thromboembolism. CONCLUSIONS: The positive predictive value of an ICD diagnosis of cancer-associated venous thromboembolism in the Danish Patient Register was overall adequate for research purposes, but with notable variation across subgroups. Sensitivity was limited, as 1/3 of patients with venous thromboembolism were discharged without any relevant ICD code. Cautious interpretation of incidence of cancer-associated venous thromboembolism based on administrative register-based data is warranted.

Somatic and psychiatric comorbidity in people with diagnosed gambling disorder: A Finnish nation-wide register study.

BACKGROUND AND AIMS: This is the first nation-wide register study based on a total population sample measuring the gender-specific incidences of chronic diseases and conditions among adults diagnosed with gambling disorder (GD). DESIGN, SETTING AND PARTICIPANTS: The study used aggregated data for 2011-22 retrieved from the Register of Primary Health Care visits, Care Register for Health Care and Care Register for Social Welfare, including specialized outpatient and inpatient health care, inpatient social care and institutional care and housing services with 24-hour or part-time assistance, set in mainland Finland. Participants comprised people aged 18-90+ years with GD diagnosis [corresponding to pathological gambling, International Classification of Diseases 10th revision (ICD-10) code F63.0, n = 3605; men n = 2574, women n = 1031] and the general population (n = 4 374 192). MEASUREMENTS: Incidences of somatic diseases and psychiatric disorders were calculated for the people with diagnosed GD and for the general population, separately for women and men. FINDINGS: After standardizing for age, the incidence of each diagnostic group was systematically higher for people with GD compared with the general population, except for cancer. The highest standardized incidence ratio (SIR) values were for psychiatric disorders [SIR = 234.2; 95% confidence interval (CI) = 226.1-242.4], memory disorders (SIR = 172.1; 95% CI = 119.1-234.8), nervous system diseases (SIR = 162.8; 95% CI = 152.8-173.1), chronic respiratory diseases (SIR = 150.6; 95% CI = 137.6-164.2), diabetes (SIR = 141.4; 95% CI = 127.9-155.5) and digestive diseases (SIR = 134.5; 95% CI = 127.1-142.2). CONCLUSIONS: In Finland, the incidence of chronic diseases and conditions among people with gambling disorder is higher compared with the general population, apart from cancer.

Evaluation of SNOMED CT Grouper Accuracy and Coverage in Organizing the Electronic Health Record Problem List by Clinical System: Observational Study.

Background: The problem list (PL) is a repository of diagnoses for patients' medical conditions and health-related issues. Unfortunately, over time, our PLs have become overloaded with duplications, conflicting entries, and no-longer-valid diagnoses. The lack of a standardized structure for review adds to the challenges of clinical use. Previously, our default electronic health record (EHR) organized the PL primarily via alphabetization, with other options available, for example, organization by clinical systems or priority settings. The system's PL was built with limited groupers, resulting in many diagnoses that were inconsistent with the expected clinical systems or not associated with any clinical systems at all. As a consequence of these limited EHR configuration options, our PL organization has poorly supported clinical use over time, particularly as the number of diagnoses on the PL has increased. Objective: We aimed to measure the accuracy of sorting PL diagnoses into PL system groupers based on Systematized Nomenclature of Medicine Clinical Terms (SNOMED CT) concept groupers implemented in our EHR. Methods: We transformed and developed 21 system- or condition-based groupers, using 1211 SNOMED CT hierarchal concepts refined with Boolean logic, to reorganize the PL in our EHR. To evaluate the clinical utility of our new groupers, we extracted all diagnoses on the PLs from a convenience sample of 50 patients with 3 or more encounters in the previous year. To provide a spectrum of clinical diagnoses, we included patients from all ages and divided them by sex in a deidentified format. Two physicians independently determined whether each diagnosis was correctly attributed to the expected clinical system grouper. Discrepancies were discussed, and if no consensus was reached, they were adjudicated by a third physician. Descriptive statistics and Cohen κ statistics for interrater reliability were calculated. Results: Our 50-patient sample had a total of 869 diagnoses (range 4-59; median 12, IQR 9-24). The reviewers initially agreed on 821 system attributions. Of the remaining 48 items, 16 required adjudication with the tie-breaking third physician. The calculated κ statistic was 0.7. The PL groupers appropriately associated diagnoses to the expected clinical system with a sensitivity of 97.6%, a specificity of 58.7%, a positive predictive value of 96.8%, and an F1-score of 0.972. Conclusions: We found that PL organization by clinical specialty or condition using SNOMED CT concept groupers accurately reflects clinical systems. Our system groupers were subsequently adopted by our vendor EHR in their foundation system for PL organization.

Cause of death coding in asthma.

BACKGROUND: While clinical coding is intended to be an objective and standardized practice, it is important to recognize that it is not entirely the case. The clinical and bureaucratic practices from event of death to a case being entered into a research dataset are important context for analysing and interpreting this data. Variation in practices can influence the accuracy of the final coded record in two different stages: the reporting of the death certificate, and the International Classification of Diseases (Version 10; ICD-10) coding of that certificate. METHODS: This study investigated 91,022 deaths recorded in the Scottish Asthma Learning Healthcare System dataset between 2000 and 2017. Asthma-related deaths were identified by the presence of any of ICD-10 codes J45 or J46, in any position. These codes were categorized either as relating to asthma attacks specifically (status asthmatic; J46) or generally to asthma diagnosis (J45). RESULTS: We found that one in every 200 deaths in this were coded as being asthma related. Less than 1% of asthma-related mortality records used both J45 and J46 ICD-10 codes as causes. Infection (predominantly pneumonia) was more commonly reported as a contributing cause of death when J45 was the primary coded cause, compared to J46, which specifically denotes asthma attacks. CONCLUSION: Further inspection of patient history can be essential to validate deaths recorded as caused by asthma, and to identify potentially mis-recorded non-asthma deaths, particularly in those with complex comorbidities.

Is complementary medicine and diagnosis associated among spinal patients in the secondary sector: a cross-sectional study.

PURPOSE: The services defined as complementary and alternative medicine/healthcare (CAM) are used to varying degrees according to the nature of the health problem, and musculoskeletal disorders, in particular, often lead to the use of CAM. Chronic pain is often cited as a reason for using CAM, and it is also the cardinal symptom of patients with back pain referred for specialist care. However, previous studies do not consider the heterogeneity of back pain when examining the use of CAM. Thus, this study aimed to explore the associations between CAM use and clinical findings incl. ICD-10 diagnostic codes in such a context. METHODS: In a cross-sectional study, a logistic regression analysis examined associations between CAM use and clinical findings at a public outpatient spine department. Chi-squared test examined the association between self-reported reasons for CAM use and the diagnostic groups. RESULTS: Of the 432 patients in the study population, 23.8% reported using CAM within 12 months prior to clinical assessment. CAM use was associated with being female and of younger age. Seeking CAM was not associated with clinical findings nor diagnosis, and no statistically significant association between the reasons for seeking CAM and the diagnostic groups was described. CONCLUSIONS: Among patients referred to specialist care for back pain, this study provides no evidence that the spinal condition should be expected to lead to the use of CAM. Only the individual demographic findings, specifically age and gender, were associated with CAM use.

Development and external validation of tools for categorizing diagnosis codes in international hospital data.

BACKGROUND: The Clinical Classification Software Refined (CCSR) is a tool that groups many thousands of International Classification of Diseases 10th Revision (ICD-10) diagnosis codes into approximately 500 clinically meaningful categories, simplifying analyses. However, CCSR was developed for use in the United States and may not work well with other country-specific ICD-10 coding systems. METHOD: We developed an algorithm for semi-automated matching of Canadian ICD-10 codes (ICD-10-CA) to CCSR categories using discharge diagnoses from adult admissions at 7 hospitals between Apr 1, 2010 and Dec 31, 2020, and manually validated the results. We then externally validated our approach using inpatient hospital encounters in Denmark from 2017 to 2018. KEY RESULTS: There were 383,972 Canadian hospital admissions with 5,186 distinct ICD-10-CA diagnosis codes and 1,855,837 Danish encounters with 4,612 ICD-10 diagnosis codes. Only 46.6% of Canadian codes and 49.4% of Danish codes could be directly categorized using the official CCSR tool. Our algorithm facilitated the mapping of 98.5% of all Canadian codes and 97.7% of Danish codes. Validation of our algorithm by clinicians demonstrated excellent accuracy (97.1% and 97.0% in Canadian and Danish data, respectively). Without our algorithm, many common conditions did not match directly to a CCSR category, such as 96.6% of hospital admissions for heart failure. CONCLUSION: The GEMINI CCSR matching algorithm (available as an open-source package at https://github.com/GEMINI-Medicine/gemini-ccsr) improves the categorization of Canadian and Danish ICD-10 codes into clinically coherent categories compared to the original CCSR tool. We expect this approach to generalize well to other countries and enable a wide range of research and quality measurement applications.

The Rise of Endovascular Repair for Abdominal, Thoracoabdominal, and Thoracic Aortic Aneurysms.

BACKGROUND: Given changes in intervention guidelines and the growing popularity of endovascular treatment for aortic aneurysms, we examined the trends in admissions and repairs of abdominal aortic aneurysms (AAA), thoracoabdominal aortic aneurysms (TAAA), and thoracic aortic aneurysms (TAA). METHODS: We identified all patients admitted with ruptured aortic aneurysms and intact aortic aneurysms repaired in the Nationwide Inpatient Sample (NIS) between 2004-2019. We then examined the utilization of open, endovascular, and complex endovascular repair (OAR,EVAR,cEVAR) for each aortic aneurysm location (AAA,TAAA,TAA), alongside their resulting in-hospital mortality, over time. cEVAR included branched, fenestrated, and physician modified endograft. RESULTS: 715,570 patients were identified with AAA (87% Intact-Repairs, 13% Rupture-Admissions). Both intact AAA repairs and ruptured AAA admissions decreased significantly between 2004 and 2019 (intact 41,060-34,215,p<.01; ruptured 7,175-4,625,p=.02). Out of all AAA repairs done in a given year, the use of EVAR increased (2004-2019: intact 45%-66%,p<.01; ruptured 10%-55%,p<.01) as well as cEVAR (2010-2019: intact 0%-23%,p<.01; ruptured 0%-14%,p<.01). Mortality after EVAR of intact AAAs decreased significantly by 29% (2004-2019, 0.73%-0.52%,p<.01) while mortality after OAR increased significantly by 16% (2004-2019, 4.4%-5.1%,p<.01). In the study, 27,443 patients were identified with TAAA (80% Intact, 20% Ruptured). In the same period, intact TAAA repairs trended upwards (2004-2019 1,435-1,640,p=.055) and cEVAR became the most common approach (2004-2019, 3.8%-72%,p=.055). 141,651 patients were identified with ascending, arch, or descending TAA (90% Intact, 10% Ruptured). Intact TAA repairs increased significantly (2004-2019 4,380-10,855,p<.01). From 2017-2019, the mortality after OAR of descending TAAs increased and mortality after TEVAR decreased (2017-2019: OAR 1.6%-3.1%; TEVAR 5.2%-3.8%). CONCLUSION: Both intact AAA repairs and ruptured AAA admissions significantly decreased between 2004 and 2019. The use of endovascular techniques for the repair of all aortic aneurysm locations, both intact and ruptured, increased over the past two decades. Most recently in 2019, 89% of intact AAAs repairs, infrarenal through suprarenal, were endovascular (EVAR or cEVAR, respectively). cEVAR alone has risen to 23% of intact AAA repairs in 2019, from 0% a decade earlier. In this period of innovation, with many new options to repair aortic aneurysms while maintaining arterial branches, endovascular repair is now used for the majority of all intact aortic aneurysm repairs. Long-term data are needed to evaluate the durability of these procedures.

Promoting interoperability between SNOMED CT and ICD-11: lessons learned from the pilot project mapping between SNOMED CT and the ICD-11 Foundation.

OBJECTIVE: To explore the feasibility and challenges of mapping between SNOMED CT and the ICD-11 Foundation in both directions, SNOMED International and the World Health Organization conducted a pilot mapping project between September 2021 and August 2022. MATERIALS AND METHODS: Phase 1 mapped ICD-11 Foundation entities from the endocrine diseases chapter, excluding malignant neoplasms, to SNOMED CT. In phase 2, SNOMED CT concepts equivalent to those covered by the ICD-11 entities in phase 1 were mapped to the ICD-11 Foundation. The goal was to identify equivalence between an ICD-11 Foundation entity and a SNOMED CT concept. Postcoordination was used for mapping to ICD-11. Each map was done twice independently, the results were compared, and discrepancies were reconciled. RESULTS: In phase 1, 59% of 637 ICD-11 Foundation entities had an exact match in SNOMED CT. In phase 2, 32% of 1893 SNOMED CT concepts had an exact match in the ICD-11 Foundation, and postcoordination added 15% of exact match. Challenges encountered included non-synonymous synonyms, mismatch in granularity, composite conditions, and residual categories. CONCLUSION: This pilot project shed light on the tremendous amount of effort required to create a map between the 2 coding systems and uncovered some common challenges. Future collaborative work between SNOMED International and WHO will likely benefit from its findings. It is recommended that the 2 organizations should clarify goals and use cases of mapping, provide adequate resources, set up a road map, and reconsider their original proposal of incorporating SNOMED CT into the ICD-11 Foundation ontology.

A Data-Driven Approach to Defining Risk-Adjusted Coding Specificity Metrics for a Large U.S. Dementia Patient Cohort.

Medical coding impacts patient care quality, payor reimbursement, and system reliability through the precision of patient information documentation. Inadequate coding specificity can have significant consequences at administrative and patient levels. Models to identify and/or enhance coding specificity practices are needed. Clinical records are not always available, complete, or homogeneous, and clinically driven metrics to assess medical practices are not logistically feasible at the population level, particularly in non-centralized healthcare delivery systems and/or for those who only have access to claims data. Data-driven approaches that incorporate all available information are needed to explore coding specificity practices. Using N = 487,775 hospitalization records of individuals diagnosed with dementia and discharged in 2022 from a large all-payor administrative claims dataset, we fitted logistic regression models using patient and facility characteristics to explain the coding specificity of principal and secondary diagnoses of dementia. A two-step approach was produced to allow for the flexible clustering of patient-level outcomes. Model outcomes were then used within a Poisson binomial model to identify facilities that over- or under-specify dementia diagnoses against healthcare industry standards across hospitalizations. The results indicate that multiple factors are significantly associated with dementia coding specificity, especially for principal diagnoses of dementia (AUC = 0.727). The practical use of this novel risk-adjusted metric is demonstrated for a sample of facilities and geospatially via a U.S. map. This study's findings provide healthcare facilities with a benchmark for assessing coding specificity practices and developing quality enhancements to align with healthcare industry standards, ultimately contributing to better patient care and healthcare system reliability.

Combining deep neural networks, a rule-based expert system and targeted manual coding for ICD-10 coding causes of death of French death certificates from 2018 to 2019.

OBJECTIVE: For ICD-10 coding causes of death in France in 2018 and 2019, predictions by deep neural networks (DNNs) are employed in addition to fully automatic batch coding by a rule-based expert system and to interactive coding by the coding team focused on certificates with a special public health interest and those for which DNNs have a low confidence index. METHODS: Supervised seq-to-seq DNNs are trained on previously coded data to ICD-10 code multiple causes and underlying causes of death. The DNNs are then used to target death certificates to be sent to the coding team and to predict multiple causes and underlying causes of death for part of the certificates. Hence, the coding campaign for 2018 and 2019 combines three modes of coding and a loop of interaction between the three. FINDINGS: In this campaign, 62% of the certificates are automatically batch coded by the expert system, 3% by the coding team, and the remainder by DNNs. Compared to a traditional campaign that would have relied on automatic batch coding and manual coding, the present campaign reaches an accuracy of 93.4% for ICD-10 coding of the underlying cause (95.6% at the European shortlist level). Some limitations (risks of under- or overestimation) appear for certain ICD categories, with the advantage of being quantifiable. CONCLUSION: The combination of the three coding methods illustrates how artificial intelligence, automated and human codings are mutually enriching. Quantified limitations on some chapters of ICD codes encourage an increase in the volume of certificates sent for manual coding from 2021 onward.

Improving Accuracy of Administrative Data for Perforated Appendicitis Classification.

INTRODUCTION: Reliance on International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) diagnosis codes may misclassify perforated appendicitis with resultant research, fiscal, and public health implications. We aimed to improve the accuracy of administrative data for perforated appendicitis classification relying on ICD-10-CM codes from 2015 to 2018. METHODS: We conducted a retrospective study of randomly sampled patients aged ≤18 years diagnosed with acute appendicitis from eight children's hospitals. Patients were identified using the Pediatric Health Information System, and true perforation status was determined by medical record review. We developed two algorithms by leveraging Pediatric Health Information System data elements and data mining (DM) approaches. The two developed algorithm performance was compared against algorithms that exclusively relied on ICD-10-CM codes using area under the curve and other measures. RESULTS: Of 1051 clinically validated encounters that were included, 383 (36.4%) patients were identified to have perforated appendicitis. The two algorithms developed using DM approaches primarily leveraged ICD-10-CM codes and length of stay. DM-developed algorithms had a significantly higher accuracy than algorithms relying exclusively on ICD-10-CM (P value < 0.01): sensitivity and specificity for DM-developed algorithms were 0.86-0.88 and 0.95-0.97, respectively, which were overall higher than algorithms that relied on only ICD-10-CM. CONCLUSIONS: This study provides an algorithm that can improve the accuracy of perforated appendicitis classification using commonly available elements in administrative data. We recommend that this algorithm is used in future appendicitis classification to ensure valid reporting, hospital-level benchmarking, and fiscal or public health assessments.

Validity of International Classification of Diseases, Tenth Revision, codes for atrial fibrillation/flutter in critically ill patients with sepsis.

BACKGROUND: Atrial fibrillation (AF) and atrial flutter (AFL) are frequently seen in critically ill sepsis patients and are associated with poor outcomes. There is a need for further research, however, studies are limited due to challenges in identifying patient cohorts. Administrative data using the International Classification of Diseases, Tenth Revision (ICD-10) are routinely used for identifying disease cohorts in large datasets. However, the validity of ICD-10 for AF/AFL remains unexplored in these populations. METHODS: This validation study included 6554 adults with sepsis and septic shock admitted to the intensive care unit. We sought to determine whether ICD-10 coding could accurately identify patients with and without AF/AFL compared to manual chart review. We also evaluated whether the date of ICD-10 code entry could distinguish prevalent from incident AF/AFL, presuming codes dated during the index admission to be incident AF/AFL. A manual chart review was performed on 400 randomly selected patients for confirmation of AF/AFL, and validity was measured using sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). RESULTS: Among the 400 randomly selected patients, 293 lacked ICD-10 codes for AF/AFL. The manual chart review confirmed the absence of AF/AFL in 286 patients (NPV 97.3%, specificity 99.7%). Among the 107 patients with ICD-10 codes for AF/AFL, 106 were confirmed to have AF/AFL by manual chart review (PPV 99.1%, sensitivity 93.0%). Out of the 114 patients with confirmed AF/AFL, 44 had ICD-10 codes dated during the index admission. All 44 were confirmed to have AF/AFL, however, 18 patients had prior documentation of AF/AFL (incident AF/AFL: PPV 59.1%). Specificity for incident (95.1%) and prevalent (99.7%) AF/AFL were high; however, sensitivity was 76.5% and 77.5%, respectively. DISCUSSION/CONCLUSION: ICD-10 codes perform well in identifying clinical AF/AFL in critically ill sepsis. However, their temporal specificity in distinguishing incidents from prevalent AF/AFL is limited.

Characterization of Post-COVID-19 Definitions and Clinical Coding Practices: Longitudinal Study.

BACKGROUND: Post-COVID-19 condition (colloquially known as "long COVID-19") characterized as postacute sequelae of SARS-CoV-2 has no universal clinical case definition. Recent efforts have focused on understanding long COVID-19 symptoms, and electronic health record (EHR) data provide a unique resource for understanding this condition. The introduction of the International Classification of Diseases, Tenth Revision (ICD-10) code U09.9 for "Post COVID-19 condition, unspecified" to identify patients with long COVID-19 has provided a method of evaluating this condition in EHRs; however, the accuracy of this code is unclear. OBJECTIVE: This study aimed to characterize the utility and accuracy of the U09.9 code across 3 health care systems-the Veterans Health Administration, the Beth Israel Deaconess Medical Center, and the University of Pittsburgh Medical Center-against patients identified with long COVID-19 via a chart review by operationalizing the World Health Organization (WHO) and Centers for Disease Control and Prevention (CDC) definitions. METHODS: Patients who were COVID-19 positive with either a U07.1 ICD-10 code or positive polymerase chain reaction test within these health care systems were identified for chart review. Among this cohort, we sampled patients based on two approaches: (1) with a U09.9 code and (2) without a U09.9 code but with a new onset long COVID-19-related ICD-10 code, which allows us to assess the sensitivity of the U09.9 code. To operationalize the long COVID-19 definition based on health agency guidelines, symptoms were grouped into a "core" cluster of 11 commonly reported symptoms among patients with long COVID-19 and an extended cluster that captured all other symptoms by disease domain. Patients having ≥2 symptoms persisting for ≥60 days that were new onset after their COVID-19 infection, with ≥1 symptom in the core cluster, were labeled as having long COVID-19 per chart review. The code's performance was compared across 3 health care systems and across different time periods of the pandemic. RESULTS: Overall, 900 patient charts were reviewed across 3 health care systems. The prevalence of long COVID-19 among the cohort with the U09.9 ICD-10 code based on the operationalized WHO definition was between 23.2% and 62.4% across these health care systems. We also evaluated a less stringent version of the WHO definition and the CDC definition and observed an increase in the prevalence of long COVID-19 at all 3 health care systems. CONCLUSIONS: This is one of the first studies to evaluate the U09.9 code against a clinical case definition for long COVID-19, as well as the first to apply this definition to EHR data using a chart review approach on a nationwide cohort across multiple health care systems. This chart review approach can be implemented at other EHR systems to further evaluate the utility and performance of the U09.9 code.

Validation Study of Cardiovascular International Statistical Classification of Diseases and Related Health Problems, Tenth Edition, Australian Modification (ICD-10-AM) Codes in Administrative Healthcare Databases (ANZACS-QI 77).

BACKGROUND: Administrative healthcare databases can be utilised for research. The accuracy of the International Statistical Classification of Diseases and Related Health Problems, Tenth Edition, Australian Modification (ICD-10-AM) coding of cardiovascular conditions in New Zealand is not known and requires validation. METHOD: International Statistical Classification of Diseases and Related Health Problems, Tenth Edition, Australian Modification coded discharges for acute coronary syndrome (ACS), heart failure (HF) and atrial fibrillation (AF), in both primary and secondary diagnostic positions, were identified from four district health boards between 1 January 2019 and 31 June 2019. A sample was randomly selected for retrospective clinician review for evidence of the coded diagnosis according to contemporary diagnostic criteria. Positive predictive values (PPVs) for ICD-10-AM coding vs clinician review were calculated. This study is also known as All of New Zealand, Acute Coronary Syndrome-Quality Improvement (ANZACS-QI) 77. RESULTS: A total of 600 cases (200 for each diagnosis, 5.0% of total identified cases) were reviewed. The PPV of ACS was 93% (95% confidence interval [CI] 89%-96%), HF was 93% (95% CI 89%-96%) and AF was 96% (95% CI 92%-98%). There were no differences in PPV between district health boards. PPV for ACS were lower in Maori vs non-Maori (72% vs 96%; p=0.004), discharge from non-Cardiology vs Cardiology services (89% vs 96%; p=0.048) and ICD-10-AM coding for unstable angina vs myocardial infarction (81% vs 95%; p=0.011). PPV for HF were higher in the primary vs secondary diagnostic position (100% vs 89%; p=0.001). CONCLUSIONS: The PPVs of ICD-10-AM coding for ACS, HF, and AF were high in this validation study. ICD-10-AM coding can be used to identify these diagnoses in administrative databases for the purposes of healthcare evaluation and research.

Diagnosis coding of neurocognitive disorders.

Objective: Diagnosis coding is a core clinical competency. A basic understanding of the structure of the International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM), the conventions and rules for diagnosis coding, and what constitutes accurate coding, is fundamental to the clinician's knowledge base. This commentary seeks to provide a practical framework for clinicians to perform accurate diagnosis coding of neurocognitive disorders. Method: This paper: (1) summarizes the structure of the ICD-10-CM, (2) describes the rules and conventions of diagnosis coding for diagnostic categories relevant to neurocognitive disorders, (3) presents clinical examples and pragmatic recommendations to help readers improve their day-to-day use of diagnosis codes, and (4) describes limitations and discrepancies in the diagnosis coding advice for neurocognitive disorders presented within the Diagnostic and Statistical Manual for Mental Disorders, Fifth Edition, Text Revision (DSM-5-TR) and the DSM-5-TR Neurocognitive Disorders Supplement. Its content originates from the ICD-10-CM itself and its companion document, the ICD-10-CM Official Guidelines for Coding and Reporting. Conclusion: The ICD-10-CM classification scheme is logically organized and easy to navigate for users who understand its structure and rules. Many neuropsychologists rely on the DSM-5-TR diagnosis coding advice, however that advice is limited with respect to the range of diagnosis codes relevant to neurocognitive disorders and their underlying causes. Relying on the ICD-10-CM directly for diagnosis coding of neurocognitive disorders, rather than the DSM-5-TR or other secondary sources, is therefore preferable and aids clinicians in accurate diagnosis coding.

Deaths Among Ambulance Patients Released from the Emergency Department Within the First 24 Hours With Nonspecific Diagnoses - Expected or Not?

BACKGROUND: Emergency patients are frequently assigned nonspecific diagnoses. Nonspecific diagnoses describe observations or symptoms and are found in chapters R and Z of the International Classification of Diseases, 10th edition (ICD-10). Patients with such diagnoses have relatively low mortality, but due to patient volume, the absolute number of deaths is substantial. However, information on cause of short-term mortality is limited. OBJECTIVES: To investigate whether death could be expected for ambulance patients brought to the emergency department (ED) after a 1-1-2 call, released with a nonspecific ICD-10 diagnosis within 24 h, and who subsequently died within 30 days. METHODS: Retrospective medical record review of adult 1-1-2 emergency ambulance patients brought to an ED in the North Denmark Region during 2017-2021. Patients were divided into three categories: unexpected death, expected death (terminal illness), and miscellaneous. Charlson Comorbidity Index (CCI) was assessed. RESULTS: We included 492 patients. Mortality was distributed as follows: Unexpected death 59.2% (n = 291), expected death (terminal illness) 25.8% (n = 127), and miscellaneous 15.0% (n = 74). Patients who died unexpectedly were old (median age of 82 years) and had CCI 1-2 (58.1%); 43.0% used at least five daily prescription drugs, and they were severely acutely ill upon arrival (24.7% with red triage, 60.1% died within 24 h). CONCLUSIONS: More than half of ambulance patients released within 24 h from the ED with nonspecific diagnoses, and who subsequently died within 30 days, died unexpectedly. One-fourth died from a pre-existing terminal illness. Patients dying unexpectedly were old, treated with polypharmacy, and often life-threateningly sick at arrival.

Development of a Method for Automatic Matching of Unstructured Medical Data to ICD-10 Codes.

Inconsistent disease coding standards in medicine create hurdles in data exchange and analysis. This paper proposes a machine learning system to address this challenge. The system automatically matches unstructured medical text (doctor notes, complaints) to ICD-10 codes. It leverages a unique architecture featuring a training layer for model development and a knowledge base that captures relationships between symptoms and diseases. Experiments using data from a large medical research center demonstrated the system's effectiveness in disease classification prediction. Logistic regression emerged as the optimal model due to its superior processing speed, achieving an accuracy of 81.07% with acceptable error rates during high-load testing. This approach offers a promising solution to improve healthcare informatics by overcoming coding standard incompatibility and automating code prediction from unstructured medical text.

Automated vs. manual coding of neuroimaging reports via natural language processing, using the international classification of diseases, tenth revision.

Objective: Natural language processing (NLP) can generate diagnoses codes from imaging reports. Meanwhile, the International Classification of Diseases (ICD-10) codes are the United States' standard for billing/coding, which enable tracking disease burden and outcomes. This cross-sectional study aimed to test feasibility of an NLP algorithm's performance and comparison to radiologists' and physicians' manual coding. Methods: Three neuroradiologists and one non-radiologist physician reviewers manually coded a randomly-selected pool of 200 craniospinal CT and MRI reports from a pool of >10,000. The NLP algorithm (Radnosis, VEEV, Inc., Minneapolis, MN) subdivided each report's Impression into "phrases", with multiple ICD-10 matches for each phrase. Only viewing the Impression, the physician reviewers selected the single best ICD-10 code for each phrase. Codes selected by the physicians and algorithm were compared for agreement. Results: The algorithm extracted the reports' Impressions into 645 phrases, each having ranked ICD-10 matches. Regarding the reviewers' selected codes, pairwise agreement was unreliable (Krippendorff α = 0.39-0.63). Using unanimous reviewer agreement as "ground truth", the algorithm's sensitivity/specificity/F2 for top 5 codes was 0.88/0.80/0.83, and for the single best code was 0.67/0.82/0.67. The engine tabulated "pertinent negatives" as negative codes for stated findings (e.g. "no intracranial hemorrhage"). The engine's matching was more specific for shorter than full-length ICD-10 codes (p = 0.00582x10-3). Conclusions: Manual coding by physician reviewers has significant variability and is time-consuming, while the NLP algorithm's top 5 diagnosis codes are relatively accurate. This preliminary work demonstrates the feasibility and potential for generating codes with reliability and consistency. Future works may include correlating diagnosis codes with clinical encounter codes to evaluate imaging's impact on, and relevance to care.